|Publication number||US7382833 B1|
|Application number||US 09/931,493|
|Publication date||Jun 3, 2008|
|Filing date||Aug 16, 2001|
|Priority date||Aug 16, 2001|
|Publication number||09931493, 931493, US 7382833 B1, US 7382833B1, US-B1-7382833, US7382833 B1, US7382833B1|
|Inventors||Maurice W. Peterson, Richard A. Groshong|
|Original Assignee||Rockwell Collins, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (34), Referenced by (10), Classifications (8), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to radio communications technology. More particularly, the present invention relates to quadrature modulators and quadrature signal processing circuits and systems.
Quadrature modulators operate on separate in-phase (I or i) and quadrature (Q or q) data corresponding to a single transmit or receive signal. Modern transmitters, such as direct conversion radio transmitters and quadrature modulators, are driven by digital circuits at very high sampling rates. Even though the in-phase and quadrature data is created digitally in perfect quadrature from the same signal, it is distorted by the analog circuits during transmission due to phase, gain and direct current (dc or DC) errors inherent in such analog circuits. As a result, compensation for these errors must be substantially continuous due to circuit variations over time, temperature and ever widening bandwidths. Therefore, an improved method of compensating for phase, gain and dc errors associated with the analog circuits of analog transmitters would be a significant improvement in the art.
A quadrature modulator compensation system for compensating the transmission of a source signal includes a transmitter which uses a transmitter input signal to generate a transmitter output. Calibration circuitry generates a phase error estimate of the transmitter as a function of an angle of intersection between an ideal transmitter envelope and an actual transmitter envelope. The calibration circuitry also generates a gain error estimate of the transmitter as a function of variation in the actual transmitter envelope, as determined by semi-major and semi-minor axes of an elliptical transmitter waveform. Further, the calibration circuitry determines a centroid of the actual transmitter envelope, and estimates dc offsets in an in-phase component and a quadrature component of the source signal as a function of the centroid of the actual transmitter envelope. Predistortion circuitry receives the phase, gain, and dc offset error estimates and the source signal as inputs and provides the transmitter input signal as a function of the phase, gain, and dc offset error estimates.
The present invention includes a method to compute the phase, gain, and dc error estimates associated with the analog circuits of quadrature transmitters. The present invention includes, and is described with reference to, an apparatus configured to implement algorithms that detect and compensate for these phase, gain, and dc offset error estimates.
A general system or block diagram is provided in
Signal source 110 is the data generator that provides data to drive the simulation for the following description which explains the methods of the present invention. In the system 100, signal source 110 provides the signal 111 to be transmitted by transmitter 130 after predistortion or correction by circuitry 120. Predistortion circuitry or function 120 changes the source data in such a way as to compensate for transmitter errors before transmission occurs. Error transmitter 130 adds errors to the signal provided by predistortion or correction circuitry 120 and translates it to the radio, or channel, frequency (rf) for transmission. Envelope detector 160 provides feedback about the transmitter's envelope to calibration circuitry or function 180 via low pass filter 170 which attenuates double channel and harmonic frequencies. The calibration algorithms or methods of the present invention implemented by calibration circuitry 180 compute correction coefficients and values which drive a correction model. Ideal receiver 140 and its associated low pass filter 150 are included for visual performance results and analysis. Detailed explanations of the functions performed by each of these blocks are provided as follows.
Signal source 110 illustrated in
s n=cos(ω·n·Ts)+j·sin(ω·n·Ts) Equation 1
The purpose of the predistortion function implemented in block 120 is to correct for phase, gain, and dc offset errors inherent in transmitter 130.
sr d [n]=T*E*E −1 *s[n]=T*s[n] Equation 2
This is the desired result after the appropriate substitutions from
K1k =GnEst k·cos(phEst k) Equation 3
K2 is a crossover term involved in phase error given by Equation 4:
K2k =GnEst k·sin(phEst k) Equation 4
S is a gain correction term to maintain unity gain given by Equation 5:
S k=(GnEst k·cos(phEst k))−1 Equation 5
Assuming that sc
The estimates of the offset errors idcEst and qdcEst are independent of the phase and gain errors and are computed as the centroid of the received envelope (from envelope detector 160). As will be described below in greater detail, the envelope of error transmitter 130 is resolved to the signal source's phase with delay compensation in order to approximate the errors as shown here.
Error transmitter 130 implements an error transmitter function which can be modeled mathematically as shown in
sr d [n]=(i c
where idc and qdc represent the actual in-phase and quadrature dc offset errors, G represents the actual gain error, and φ represents the actual phase error inherent in the transmitter (these are the errors to be estimated by calibration circuitry 180).
Envelope detector 160 in
The calibration circuitry function 180 shown in
As can be seen in
Sample delay circuitry 410 receives signal s[n] from signal source 110, as well as delay, dly, to compensate for pure time (i.e., sample) delay when needed. Dly is a pure time delay representing the time compensation required to allow the association of the proper samples when computing coordinates of intersection. It is an artifact of hardware and filtering delays from the signal source to the calibration function that can be determined off-line and stored in a table or constant. The output ss[n] 411 can be defined for example using the relationship shown in Equation 9.
ss n =s n−dly Equation 9
Envelope detector 415 is used to detect the quadrature signal Sr[n] 406 formed during the resolution of the filtered envelope Efr[n] 171 to the source's phase and compensation for delay. This function may not be needed depending on sampling resolution, timing and phase delays. The real and imaginary operators are a modeling convenience. Any implementation will normally use rectangular coordinates. The envelope output 416, Envr[n], of envelope detector 415 is given by:
Re(Srm) is the real portion of Sr[n]; and
Im(Srn) is the imaginary portion of Sr[n].
Envelope detector 420 is used to compute the envelope of source signal S[n] 111, including any necessary sample delay θdly, which will be used as the desired or reference waveform. The mathematical model can be the same as shown in Equation 10, with the appropriate substitutions made at the input and output (ss[n] for Sr[n], and Envt[n] for Envr[n] respectively).
Intersection identifying circuitry 425 finds the points of intersection of the received and filtered transmitter envelope, Envr[n] 416, with the desired waveform, Envt[n] 421. Index of intersections ix(ix[k] in
Slope identifying circuitry 430 finds the slope of each intersecting curve sr[n] and ss[n] at the point of intersection (n=ix). Any intersecting point in one phasor revolution is sufficient. This method uses the intersecting point sample Sr[ix] (intersecting point in curve Sr[n]) and SS[ix−1] (intersecting point in curve SS[n]) and the previous sample in the approximation of the slope, but linear interpolation could be used to find two points on either side of the intersecting point on the respective curves. This latter method could be used to give a more accurate slope value at the point of intersection between the curves and to possibly reduce the required sampling rate. The complex outputs mrc[ix] 432 and mtc[ix] 433 of slope identifying function 430 can be determined using the relationships of Equations 13 and 14, which are modeled in
mrc ix =Re(s r
Srix=Sr[n] for n=ix (point of intersection) and;
Srix−1=Sr[n] for n=ix−1 (previous sample).
mt ix =Re(ss ix)−Re(ss ix−1)+j·(Im(ss ix)−Im(ss ix−1)) Equation 14
SSix=SS[n] for n=ix; and
SSix−1=SS[n] for n=ix−1 (previous sample).
The phase error of the transmitter expresses itself in the angle of intersection of the desired and actual transmitter envelopes Envt[n] and Envr[n], respectively. As shown in
It can be shown that:
tan(α)=tan(β1−β1)=(tan(β2)−tan(β1))·(1+tan(β2)·tan(β1))−1 Equation 15
Now, the tangent value effectively gives the slope of the line 620 through the point of intersection 625, and we will let mr represent tan(β2) & mt represent tan(β1) giving:
Where INV is discussed below and where the slopes are formed as shown in Equations 17 and 18 (division by zero must be avoided).
mt k =Im(mtc ix)·(Re(mtc ix)−1 Equation 17
mr k =Im(mrc ix)·(Re(mrc ix))−1 Equation 18
Therefore, assuming that the angle of intersection α of the two waveforms approximates the transmitter phase error, then the slope of the intersections can be computed using the above equation thus giving a method of estimating the phase error. The INV factor in the Equation 16 is needed because there is a phase ambiguity depending upon which intersection point is used. If the magnitude of the imaginary coordinate of the point of intersection 625 is greater than the magnitude of the real part, INV=−1, otherwise it is 1.
A circuit 435 for implementing the mathematics described above is shown in
Referring back to
Calibration circuitry 180 also includes gain estimate computation circuitry 445. Circuitry 445 receives as inputs from circuitry 440 the semi-major and semi-minor axes Ex[k] 441 and En[k] 442. Circuit 445 integrates the envelope variation with the old gain estimate after the phase is toggled for each iteration. The phase inversion which may be replaced by a 1-pole averager is represented as EvenOdd which alternately inverts the sign of the envelope variation and is needed to maintain stability in the gain estimates because the phase estimate can be occurring simultaneously and the two estimates interact. The gain estimate GnEst[k] 182 provided by circuitry 445 can converge to within 0.01 db of the actual gain G. It can be filtered (not shown) to provide noise immunity similar to the handling of the Phase Estimate mentioned above. The output is given by Equation 19:
GnEst k =GnEs k−1+EvenOdd·(Ex k −En k) Equation 19
Also included in calibration circuitry 180 are centroid computation circuitry 450 and dc estimate generating circuitry 455. Circuitry 450 implements a mathematical model for computing the centroid of the filtered transmitter envelope. The components of the Calculus of the centroid of a curve include the length, L, of the curve and the moments of each element of L with respect to each of two aces (I and Q in this case). The moment of each element of the waveform is computed for both the in-phase and quadrature components of the envelope. The length L, of the path traced out by the received and filtered, transmitter envelope phasor, sr (406), is also computed. L is used in the next two computations of the centroid of the phasor path, idc and qdc. Other methods of effectively computing the centroid by using the envelope error, Enve, may be possible. The resulting updates are then integrated by the dc estimate computation circuitry 455 with the previous dc offset estimates for use during the next epoch or decimation period. The envelope notation used here, sr (406), results from the quadrature resolution of Efr[n] 171. The number of samples in a rotation of the envelope phasor is 0.5*nx where nx is the total number of samples generated for study (two rotations, or 4*pi, worth of samples). The last half (thus the 0.5 factor on nx below) are used to avoid transient effects in the data.
The waveform path length L is then given by Equation 20:
where nx is defined above and sr[b] and sr[b−1] (406) are adjacent samples of the quadrature resolution of Efr (171).
The quadrature components of the centroid of the received and filtered envelope waveform can then be computed as shown in Equations 21 and 22 which effectively satisfy the evaluation of the well known, however more recognizable, equations for the centrod of a curve in an x-y plane (xc=ƒx*ds and yc=ƒy*ds) in discrete form:
The centroid coordinates are integrated with the appropriate previous dc error estimates. The dc estimates also will be filtered (not shown) to provide noise immunity similar to the handling of the phase and gain estimates mentioned above. The outputs 183 and 184 can be given by Equations 23 and 24:
The invention has been described as generating a phase error estimate, a gain error estimate, and a dc offset estimate. Said estimates may be generated simultaneously or in any order. Furthermore, any combination of said estimates may be performed as desired.
Although the present invention has been described with reference to illustrative embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the invention.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3845390 *||Nov 21, 1972||Oct 29, 1974||Philips Corp||System for automatic equalization|
|US5012208 *||Apr 10, 1990||Apr 30, 1991||Telenokia Oy||Quadrature modulator having compensation for local oscillator leak|
|US5023937 *||May 9, 1989||Jun 11, 1991||Motorola, Inc.||Transmitter with improved linear amplifier control|
|US5249204 *||Aug 12, 1991||Sep 28, 1993||Motorola, Inc.||Circuit and method for phase error correction in a digital receiver|
|US5321726 *||Nov 13, 1992||Jun 14, 1994||Hewlett-Packard Company||Calibration of vector demodulator using statistical analysis|
|US5506546 *||Jun 15, 1995||Apr 9, 1996||Nec Corporation||Method and apparatus for generating transmitting wave|
|US5507014 *||Feb 25, 1993||Apr 9, 1996||Motorola, Inc.||Radio transmitter circuit with time dependant inhibited transmissions|
|US5675287 *||Feb 12, 1996||Oct 7, 1997||Motorola, Inc.||Digital DC correction circuit for a linear transmitter|
|US5705958 *||Sep 26, 1996||Jan 6, 1998||Alcatel Telspace||Apparatus for correcting quadrature error in a modulator and/or in a demodulator for a signal having a plurality of phase states, a corresponding transmitter, and a corresponding receiver|
|US5732333 *||Feb 14, 1996||Mar 24, 1998||Glenayre Electronics, Inc.||Linear transmitter using predistortion|
|US5745006 *||Nov 12, 1996||Apr 28, 1998||Motorola, Inc.||Method of compensating for distortion in an amplifier|
|US5937011 *||Mar 26, 1996||Aug 10, 1999||Airnet Communications Corp.||Multi-carrier high power amplifier using digital pre-distortion|
|US6141377 *||Jul 30, 1998||Oct 31, 2000||Cisco Technology, Inc.||Method and apparatus for voice frequency noise reduction in splitterless ADSL|
|US6141390 *||Nov 6, 1997||Oct 31, 2000||Glenayre Electronics, Inc.||Predistortion in a linear transmitter using orthogonal kernels|
|US6211733 *||Oct 22, 1999||Apr 3, 2001||Powerwave Technologies, Inc.||Predistortion compensation for a power amplifier|
|US6232835 *||Feb 12, 1999||May 15, 2001||Nortel Networks Limited||System and method of linearizing the gain error of a power amplifier|
|US6252912 *||Dec 24, 1997||Jun 26, 2001||General Dynamics Government Systems Corporation||Adaptive predistortion system|
|US6266517 *||Dec 30, 1999||Jul 24, 2001||Motorola, Inc.||Method and apparatus for correcting distortion in a transmitter|
|US6275685 *||Dec 10, 1998||Aug 14, 2001||Nortel Networks Limited||Linear amplifier arrangement|
|US6288610 *||Jun 2, 1999||Sep 11, 2001||Fujitsu Limited||Method and apparatus for correcting signals, apparatus for compensating for distortion, apparatus for preparing distortion compensating data, and transmitter|
|US6313703 *||Mar 2, 2000||Nov 6, 2001||Datum Telegraphic, Inc||Use of antiphase signals for predistortion training within an amplifier system|
|US6314142 *||Jun 17, 1997||Nov 6, 2001||Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V.||Pre-distortion for a non-linear transmission path in the high frequency range|
|US6373902 *||Jun 19, 1998||Apr 16, 2002||Samsung Electronics Co., Ltd||Device and method for linearizing transmitter in digital communication system|
|US6591090 *||May 24, 1999||Jul 8, 2003||Nokia Mobile Phones Limited||Predistortion control for power reduction|
|US6654426 *||Sep 26, 2001||Nov 25, 2003||Nokia Networks Oy||Correction of nonlinearity of I/Q modulator|
|US6674808 *||Dec 28, 1999||Jan 6, 2004||General Dynamics Decision Systems, Inc.||Post-amplifier filter rejection equalization|
|US6711217 *||Jul 19, 1999||Mar 23, 2004||Samsung Electronics Co., Ltd.||Apparatus and method for linearized power amplification|
|US6720839 *||Feb 29, 2000||Apr 13, 2004||Skyworks Solutions, Inc.||System of and method for reducing or eliminating the unwanted sideband in a signal derived from the output of a quadrature modulator|
|US6741662 *||Apr 17, 2000||May 25, 2004||Intel Corporation||Transmitter linearization using fast predistortion|
|US6756845 *||Jan 26, 2000||Jun 29, 2004||Nokia Corporation||Method and system for compensating non-linearities and time-varying changes of a transfer function acting on an input signal|
|US7062233 *||Jun 14, 2001||Jun 13, 2006||Nokia Corporation||Transmitter linearization|
|US20010014592 *||Jul 8, 1998||Aug 16, 2001||Jochen Helms||Transmitter|
|US20020016154 *||Jun 14, 2001||Feb 7, 2002||Mikko Huttunen||Transmitter linearization|
|US20020034260 *||Sep 14, 2001||Mar 21, 2002||Lg Electronics Inc.||Adaptive predistortion transmitter|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7627055 *||Apr 8, 2003||Dec 1, 2009||Nokia Corporation||Error adjustment in direct conversion architectures|
|US7831220 *||Apr 4, 2006||Nov 9, 2010||Broadcom Corporation||Methods and systems for calibrating for gain and phase imbalance and local oscillator feed-through|
|US7893787 *||Aug 19, 2005||Feb 22, 2011||Nec Corporation||DC offset cancellation circuit for modulator using 1-bit signal conversion|
|US8290450||Nov 4, 2010||Oct 16, 2012||Broadcom Corporation||Methods and systems for calibrating for gain and phase imbalance and local oscillator feed-through|
|US20040193965 *||Apr 8, 2003||Sep 30, 2004||Edmund Coersmeier||Error adjustment in direct conversion architectures|
|US20070025433 *||Apr 4, 2006||Feb 1, 2007||Broadcom Corporation||Methods and systems for calibrating for gain and phase imbalance and local oscillator feed-through|
|US20080043881 *||Jun 14, 2007||Feb 21, 2008||Reiner Franke||System for calibrating at least one quadrature modulator and operating method therefor|
|US20090261918 *||Aug 19, 2005||Oct 22, 2009||Nec Corporation||Dc offset cancellation circuit for modulator using 1-bit signal conversion|
|US20110051791 *||Nov 4, 2010||Mar 3, 2011||Broadcom Corporation||Methods and Systems for Calibrating for Gain and Phase Imbalance and Local Oscillator Feed-Through|
|EP2779553A3 *||Mar 6, 2014||Dec 10, 2014||Analog Devices Technology||DC bias estimation of a radio frequency mixer|
|U.S. Classification||375/296, 375/285|
|Cooperative Classification||H04L2027/0016, H04L27/364, H04L27/368|
|European Classification||H04L27/36G1A, H04L27/36B3|
|Aug 16, 2001||AS||Assignment|
Owner name: ROCKWELL COLLINS, INC., IOWA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:PETERSON, MAURICE W.;GROSHONG, RICHARD A.;REEL/FRAME:012104/0983
Effective date: 20010813
|Sep 23, 2011||FPAY||Fee payment|
Year of fee payment: 4
|Dec 3, 2015||FPAY||Fee payment|
Year of fee payment: 8